LUNA Project Comes to a Successful End

Loquendo last week successfully completed the three-year LUNA project (Spoken Languange UNderstanding in MultilinguAl Communication Systems), a European Union-sponsored initiative to address the problem of understanding spontaneous speech in the context of automated telephone systems. 

The project aimed at developing robust technology capable of understanding customers' spoken requests, to allow the deployment of advanced vocal services, and to enable callers to interact by means of spoken language understanding (SLU).

The ambitious goal of the project was to position itself at the forefront of the third generation of spoken language interfaces, by proposing new methods, algorithms, and tools to solve complex tasks and adapt to the context in which the dialogue is taking place. A three-tier software architecture has been used for SLU. The tiers implement the following processes: generation of semantic concept tags, semantic composition, and context-sensitive validation.

The project used real data collected in human-human and human-machine complex dialogues in three languages. The results were validated with prototypes applied to different tasks: a tourism application in French, a help desk service in Italian, and a route planner for city public transport in Polish.

This technology offers advantages in terms of the evolution of human-machine interaction in automated call centers, resulting in reduced costs for service providers and an enhanced user experience. It allows an increase in the automation rate of telephone services, permitting a better distribution of the work load between operator and machine, and less waiting time for callers. Moreover, as customers learn to use the system, interaction becomes even more rapid and effective.

The LUNA research results will be applicable to many other sectors, besides telephony and automated call centers.

"Loquendo is proud of this result" said Silvia Mosso, LUNA project coordinator. "We have successfully managed a complex and challenging project, working with leading research institutes and industrial partners in Europe. The results we have achieved demonstrate the excellence of the consortium and pave the way for new and more ambitious research topics. We think that in the near future we will see an ever-increasing pervasiveness of speech interfaces, which will become widespread, not just in call center applications, but for accessing any information by voice. Loquendo will continue to invest in research and development, to offer to its customers the best-of-breed speech technology."

"In the LUNA project we have accomplished three important results. We have designed machine learning algorithms and set the state-of-the-art for robust spoken language understanding for multiple languages and different complexity applications. We have proposed potentially powerful semantic models for understanding language in context which could support next-generation problem solving conversational machines" said Giuseppe Riccardi, professor and head of the Adaptive Multimodal Information and Interface Lab at the University of Trento "Last but not least, our industry and public administration partners have gained insights in such technology and evaluated LUNA dialogue system prototypes that we hope to see in the speech market soon ."